Twitter 'offers world mood clues'

Image caption The researchers say Twitter provides a real-time reflection of mood.

Researchers have trawled through more than half a billion messages on Twitter, looking for changes in people's mood across days and seasons.

They said people start off brightly in the morning, but mood deteriorates throughout the day until a boost late in the evening.

Unsurprisingly, they found people tended to be happier on Saturdays and Sundays than during the working week.

Changes in the hours of daylight also had an effect.

It has been difficult for scientists to investigate changes in mood. The authors of this study argue that traditional laboratory tests are limited as they are based on university students self-reporting their mood.

Some academics have begun looking at social networking sites to gather data on a wider and more diverse group of people.

Mood rhythm

The team at Cornell University, New York, argued that monitoring Twitter provides a real-time reflection of mood.

They looked at messages from 2.4 million people's public tweets from 84 countries and used language detection software to score positive and negative feelings.

Positive feelings started high, but began to fall in mid-morning and then picked up again in the evening. Generally, positive words appeared more at the weekend than during the week.

However, the researchers say work is not necessarily to blame as the shift in mood pattern "was similar on weekends and weekdays" which instead "points to sleep and the biological clock" as the cause.

Negative feelings were lowest in the morning and increased throughout the day. The researchers said: "This pattern also suggests that people may be emotionally refreshed by sleep."

Researchers said day length affected positive tweets, but not negative ones suggesting "that winter blues is associated with diminished positive affect but not increased negative affect".

More on this story

Related Internet links

The BBC is not responsible for the content of external Internet sites